That would be an intentional poisoning of the models with biased or outright untruthful data.
I believe that many people would be unwilling to use such models.